Right on!
The checklist is on its way – it should land in your inbox in a few minutes!
(And don’t forget to check your spam folder, just in case!)
How we crafted personalized offline event invites at scale for cold prospects
Tools: Clay
Enrichment: Icypeas, GPT 4.1 Nano, GPT 4.1 Mini, GPT 5
One of my clients was preparing for an event and asked me to build a personalized Clay workflow - repeatable and scalable for other events - to send personalized invitations to their prospects.
We had 4K prospects with valid emails, and our goal was to create a campaign to personally invite all of them to our offline e-commerce event in Manchester for UK e-commerce stores.
When you invite someone to an event, the best email should explain why that person should come. It’s still a cold email, so it’s a bad idea to throw in the entire agenda, ticket costs, and all the details - that will likely get you sent straight to spam.
The email should answer one question: what’s in this event that might be valuable for me?
Usually, when people attend an event, it’s because of:
● Networking
● A speaker covering a topic they’re interested in
● A competitor speaking about how they solved a specific problem
For 40% of the list, we had LinkedIn profiles, so I suggested scraping the About section to understand the person better, then asking AI to match the description with the event agenda - suggesting the most relevant speaker based on their LinkedIn description.
For the rest, we had industry data, so we simply matched talks to their industries - assuming, for example, that the CEO of an apparel store is more likely to be interested in a talk from another apparel CEO.
So our email structure looked like this:
Hi John,
Opener - why we’re reaching out (because we noticed the company among UK stores)
Why we think the event is relevant based on prospect and company info
CTA - asking if the prospect is interested in getting more details
1. Generating the opener with Claygent
To make the email feel manual, we don’t use any generic lines - instead, we mention the type of e-commerce store, not just the broad category (which is often wrong and can ruin outreach).
So the opener might be:
“Noticed [Store Name] standing out among fast-scaling [specialization] stores in the UK.”
It can also be tricky to mention “growing” - usually, “growth” can be inferred from job openings, traffic increases, or reported revenue growth.
However, our offer wasn’t directly connected to growth, so we skipped that.
(Note: When I’m writing this article, I think it could also be nice to use this angle to highlight the event’s value depending on their stage - if they’re growing, one angle; if they’ve plateaued, another.)
So that’s the prompt we used and the result:
2. Scraping the about section with Icypeas
We used Icypeas integration to scrape the About section from each person’s LinkedIn profile. We didn’t ask Claygent to visit every profile, as that would take too long and be too expensive.
When we get data from Icypeas, it’s not just the About section - it’s the entire profile. So we can process that using GPT, which is much cheaper.
3. Generating personalized openers
If a person has an About section and it’s not empty, we create a personal opener. To make GPT work based on the agenda, we uploaded the agenda document to the context section.
The results looked as if I personally went to their LinkedIn profile and checked what they do.
If there wasn’t a specific topic connected to the person’s description, AI made a general suggestion based on the agenda document (which was comprehensive enough).
4. Creating openers based on industry
For the rest of the people (without LinkedIn data), we created openers based on their industry.
Result:
5. Follow-up with a new angle
Then we created a follow-up with a slightly different angle about the event, based on the first email. I don’t like “hey, just following up” messages - every step should add new value or a new angle.
So we added more context using GPT and the previously selected speaker or angle.
6. Location-based personalization
It was an offline event, but the client also had a video stream option - so people who couldn’t attend in person could join online.
For each prospect, we had their location - and of course, it made sense to offer the video stream if the person lived far away.
We added an enrichment to calculate driving time from their location to Manchester and used it for personalization. We consulted with our clients about what people in the UK consider a “long drive.”
If it’s more than 2 hours from Manchester, we offer the video stream. If it’s less, we say something like:
“Hey, it’s just 40 minutes to Manchester - join us!”
If they’re from Manchester, we send the venue location directly.
Final email looked like this:
Explore more topics